Model profile
Glm 4 9B Chat
Z.aideveloper
2024-06-05release date
#152 / 267overall rank
7eval lineages
Evidence summary
Glm 4 9B Chat has an estimated overall rank of #152; its 90% source-sensitivity interval is #69–#235. Its behavior-only rank is #150; company governance moves the combined estimate to #152. Published evidence spans 7 evals and 4 of 7 behavior components. Its strongest relative result is SafeDialBench (ethics, #1 of 18); its weakest is Agent-SafetyBench (spread_unsafe_information, #16 of 16).
Compare this model
Only models sharing at least one published sub-eval are listed.
Official and reference links
- Hugging Face ↗zai-org/glm-4-9b-chat
- Official model page ↗Exact model document · Reviewed official Hugging Face owner · official repository
- Release source ↗direct research preview identity date
Published eval results
Rank is within that sub-eval. Black marks the observed result; the grey dot marks the value implied by the global rank. Values stay on each source’s native scale.
| Eval / sub-eval | Rank | Value | Distribution | Better | Source |
|---|---|---|---|---|---|
| Agent-SafetyBenchcompromise_availability | #1 / 16 | 50.8 | ↑ higher | Source ↗official | |
| Agent-SafetyBenchharmful_vulnerable_code | #15 / 16 | 23.2 | ↑ higher | Source ↗official | |
| Agent-SafetyBenchleak_sensitive_information | #6 / 16 | 38.4 | ↑ higher | Source ↗official | |
| Agent-SafetyBenchphysical_harm | #5 / 16 | 41.6 | ↑ higher | Source ↗official | |
| Agent-SafetyBenchproduce_unsafe_information | #16 / 16 | 57.2 | ↑ higher | Source ↗official | |
| Agent-SafetyBenchproperty_loss | #4 / 16 | 48 | ↑ higher | Source ↗official | |
| Agent-SafetyBenchspread_unsafe_information | #16 / 16 | 6 | ↑ higher | Source ↗official | |
| Agent-SafetyBenchviolate_law_ethics | #8 / 16 | 27.2 | ↑ higher | Source ↗official | |
| ChineseSafescore | #16 / 22 | 50.03 | ↑ higher | Source ↗official | |
| DSPSafeBenchscore | #7 / 12 | 72.43 | ↑ higher | Source ↗official | |
| Enkrypt AI Safety Leaderboardbias_attack_non_success_rate | #105 / 260 | 18.35 | ↑ higher | Source ↗official | |
| Enkrypt AI Safety Leaderboardcbrn_attack_non_success_rate | #204 / 260 | 82.33 | ↑ higher | Source ↗official | |
| Enkrypt AI Safety Leaderboardharmful_attack_non_success_rate | #198 / 260 | 48.33 | ↑ higher | Source ↗official | |
| Enkrypt AI Safety Leaderboardtoxicity_attack_non_success_rate | #245 / 258 | 75.41 | ↑ higher | Source ↗official | |
| JailBenchjailbreak_success_rate | #13 / 14 | 74.95 | ↓ lower | Source ↗official | |
| M3-SafetyBenchoverall_score | #15 / 19 | 84.2 | ↑ higher | Source ↗official | |
| SafeDialBenchaggression | #6 / 18 | 7.17 | ↑ higher | Source ↗official | |
| SafeDialBenchethics | #1 / 18 | 7.753 | ↑ higher | Source ↗official | |
| SafeDialBenchfairness | #8 / 18 | 7.4 | ↑ higher | Source ↗official | |
| SafeDialBenchlegality | #2 / 18 | 8.103 | ↑ higher | Source ↗official | |
| SafeDialBenchmorality | #3 / 18 | 7.427 | ↑ higher | Source ↗official | |
| SafeDialBenchprivacy | #2 / 18 | 7.853 | ↑ higher | Source ↗official |